agentvibes 5.12.0 → 5.13.0

package/src/services/provider-voice-catalog.js

@@ -1,87 +1,55 @@
 /**
- * Provider voice catalogs — single source of truth for the static voice lists
- * of the cloud/neural providers, plus a helper that returns the voice pool for
- * whichever provider is currently active.
+ * Provider voice catalogs — BACK-COMPAT SHIM.
  *
- * Piper voices are NOT listed here (they are discovered on disk via
- * scanInstalledVoices); this module covers the providers whose voices are a
- * fixed catalog: ElevenLabs and Kokoro. Single-voice providers (e.g. soprano)
- * resolve to a single synthetic entry.
+ * The canonical source of provider voice data is now
+ * `src/services/provider-catalog.js` (the Provider Catalog, SSOT Layer 2). This
+ * module re-exports the static voice lists (`ELEVENLABS_VOICES`,
+ * `KOKORO_VOICE_IDS`) and the gender helper from the catalog, and keeps the
+ * `voicesForProvider()` convenience used by BMAD gender-aware auto-assignment.
+ *
+ * Kept as a shim so existing importers (and test/unit/provider-voice-catalog.test.js)
+ * continue to work unmodified.
  *
  * @module services/provider-voice-catalog
  */
 
 import { isSingleVoiceProvider } from './agent-voice-store.js';
+import {
+  ELEVENLABS_VOICES,
+  KOKORO_VOICE_IDS,
+  kokoroGender,
+} from './provider-catalog.js';
+
+export { ELEVENLABS_VOICES, KOKORO_VOICE_IDS, kokoroGender };
 
-/** ElevenLabs library voices (voice_id + display metadata). */
-export const ELEVENLABS_VOICES = [
-  { id: 'EXAVITQu4vr4xnSDxMaL', name: 'Sarah',   gender: 'Female', lang: 'en-US', desc: 'Mature, reassuring'      },
-  { id: 'CwhRBWXzGAHq8TQ4Fs17', name: 'Roger',   gender: 'Male',   lang: 'en-US', desc: 'Laid-back, casual'       },
-  { id: 'FGY2WhTYpPnrIDTdsKH5', name: 'Laura',   gender: 'Female', lang: 'en-US', desc: 'Enthusiast, quirky'      },
-  { id: 'IKne3meq5aSn9XLyUdCD', name: 'Charlie', gender: 'Male',   lang: 'en-AU', desc: 'Deep, confident'         },
-  { id: 'JBFqnCBsd6RMkjVDRZzb', name: 'George',  gender: 'Male',   lang: 'en-GB', desc: 'Warm storyteller'        },
-  { id: 'N2lVS1w4EtoT3dr4eOWO', name: 'Callum',  gender: 'Male',   lang: 'en-US', desc: 'Husky trickster'         },
-  { id: 'SAz9YHcvj6GT2YYXdXww', name: 'River',   gender: '',       lang: 'en-US', desc: 'Relaxed, neutral'        },
-  { id: 'SOYHLrjzK2X1ezoPC6cr', name: 'Harry',   gender: 'Male',   lang: 'en-US', desc: 'Fierce warrior'          },
-  { id: 'TX3LPaxmHKxFdv7VOQHJ', name: 'Liam',    gender: 'Male',   lang: 'en-US', desc: 'Energetic creator'       },
-  { id: 'Xb7hH8MSUJpSbSDYk0k2', name: 'Alice',   gender: 'Female', lang: 'en-GB', desc: 'Clear educator'          },
-  { id: 'XrExE9yKIg1WjnnlVkGX', name: 'Matilda', gender: 'Female', lang: 'en-US', desc: 'Knowledgable, pro'       },
-  { id: 'bIHbv24MWmeRgasZH58o', name: 'Will',    gender: 'Male',   lang: 'en-US', desc: 'Relaxed optimist'        },
-  { id: 'cgSgspJ2msm6clMCkdW9', name: 'Jessica', gender: 'Female', lang: 'en-US', desc: 'Playful, bright'         },
-  { id: 'cjVigY5qzO86Huf0OWal', name: 'Eric',    gender: 'Male',   lang: 'en-US', desc: 'Smooth, trustworthy'     },
-  { id: 'hpp4J3VqNfWAUOO0d1Us', name: 'Bella',   gender: 'Female', lang: 'en-US', desc: 'Professional, warm'      },
-  { id: 'iP95p4xoKVk53GoZ742B', name: 'Chris',   gender: 'Male',   lang: 'en-US', desc: 'Charming, down-to-earth' },
-  { id: 'nPczCjzI2devNBz1zQrb', name: 'Brian',   gender: 'Male',   lang: 'en-US', desc: 'Deep, comforting'        },
-  { id: 'onwK4e9ZLuTAKqWW03F9', name: 'Daniel',  gender: 'Male',   lang: 'en-GB', desc: 'Steady broadcaster'      },
-  { id: 'pFZP5JQG7iQjIQuC4Bku', name: 'Lily',    gender: 'Female', lang: 'en-GB', desc: 'Velvety actress'         },
-  { id: 'pNInz6obpgDQGcFmaJgB', name: 'Adam',    gender: 'Male',   lang: 'en-US', desc: 'Dominant, firm'          },
-  { id: 'pqHfZKP75CvOlQylNhV4', name: 'Bill',    gender: 'Male',   lang: 'en-US', desc: 'Wise, mature'            },
-  // ── Added from the ElevenLabs voice library (raw library voice_ids) ───────
-  { id: 'HBDoL4wkcalemIO0nUAu', name: 'Emily',   gender: 'Female', lang: 'en-GB', desc: 'RPG, immersive'    },
-  { id: 'Mtmp3KhFIjYpWYRycDe3', name: 'John',    gender: 'Male',   lang: 'en-US', desc: 'Raspy surfer'      },
-  { id: '8kgj5469z1URcH4MB2G4', name: 'Sakuya',  gender: 'Female', lang: 'en-US', desc: 'Cheerful anime'    },
-  { id: '2ajXGJNYBR0iNHpS4VZb', name: 'Rob',     gender: 'Male',   lang: 'en-GB', desc: 'Tough, gritty'     },
-  { id: 'mMdTuD2nnFiCH88UdXSb', name: 'Jane',    gender: 'Female', lang: 'en-US', desc: 'Cinematic villain' },
+/**
+ * Curated built-in voices for the OS-native "discovered" providers (Windows SAPI,
+ * macOS `say`). Their real voices live on the target device — for a remote
+ * receiver we cannot enumerate them from here — so these standard voices (shipped
+ * by default on Windows 10/11 and macOS) give the picker an honest, selectable
+ * list. The `id` is the exact name the player passes to SelectVoice / `say -v`;
+ * if the receiver lacks one it falls back to the system default (benign).
+ *
+ * Verified against a real Windows 11 install (System.Speech GetInstalledVoices).
+ * The receiver remains authoritative — this is UI convenience, not inventory SSOT.
+ */
+const WINDOWS_SAPI_VOICES = [
+  { id: 'Microsoft David Desktop', name: 'David (Desktop)', gender: 'Male',   lang: 'en-US' },
+  { id: 'Microsoft Zira Desktop',  name: 'Zira (Desktop)',  gender: 'Female', lang: 'en-US' },
+  { id: 'Microsoft David',         name: 'David',           gender: 'Male',   lang: 'en-US' },
+  { id: 'Microsoft Mark',          name: 'Mark',            gender: 'Male',   lang: 'en-US' },
+  { id: 'Microsoft Zira',          name: 'Zira',            gender: 'Female', lang: 'en-US' },
 ];
 
-/** Kokoro voice ids. Gender is encoded in the id: the 2nd char is f|m. */
-export const KOKORO_VOICE_IDS = [
-  // American English
-  'af_heart','af_alloy','af_aoede','af_bella','af_jessica','af_kore','af_nicole','af_nova','af_river','af_sarah','af_sky',
-  'am_adam','am_echo','am_eric','am_fenrir','am_liam','am_michael','am_onyx','am_puck',
-  // British English
-  'bf_alice','bf_emma','bf_isabella','bf_lily',
-  'bm_daniel','bm_fable','bm_george','bm_lewis',
-  // Japanese
-  'jf_alpha','jf_gongitsune','jf_nezumi','jf_tebukuro','jm_kumo',
-  // Mandarin Chinese
-  'zf_xiaobei','zf_xiaoni','zf_xiaoxiao','zf_xiaoyi','zm_yunxi','zm_yunxia','zm_yunyang',
-  // Spanish
-  'ef_dora','em_alex','em_santa',
-  // French
-  'ff_siwis',
-  // Hindi
-  'hf_alpha','hm_omega',
-  // Italian
-  'if_sara','im_nicola',
-  // Brazilian Portuguese
-  'pf_dora','pm_alex','pm_santa',
-  // Korean
-  'kf_alpha','km_hyunsu',
+const MACOS_VOICES = [
+  { id: 'Samantha', name: 'Samantha', gender: 'Female', lang: 'en-US' },
+  { id: 'Alex',     name: 'Alex',     gender: 'Male',   lang: 'en-US' },
+  { id: 'Daniel',   name: 'Daniel',   gender: 'Male',   lang: 'en-GB' },
+  { id: 'Karen',    name: 'Karen',    gender: 'Female', lang: 'en-AU' },
+  { id: 'Moira',    name: 'Moira',    gender: 'Female', lang: 'en-IE' },
 ];
 
-/**
- * Infer a Kokoro voice's gender from its id. Kokoro ids follow `<lang><sex>_name`
- * where the 2nd character is `f` (female) or `m` (male).
- * @param {string} id
- * @returns {'Female'|'Male'|''}
- */
-export function kokoroGender(id) {
-  const c = typeof id === 'string' && id.length > 1 ? id[1].toLowerCase() : '';
-  if (c === 'f') return 'Female';
-  if (c === 'm') return 'Male';
-  return '';
-}
+export { WINDOWS_SAPI_VOICES, MACOS_VOICES };
 
 /**
  * Return the voice pool for the currently-active provider as `{ id, gender }`
@@ -108,6 +76,17 @@ export function voicesForProvider(provider, { scanInstalledVoices, getVoiceMeta
     return ELEVENLABS_VOICES.map(v => ({ id: v.id, gender: v.gender || '' }));
   }
 
+  // OS-native discovered providers: their voices live on the target device (the
+  // receiver in remote mode), so they can't be scanned here. Return the curated
+  // built-in list instead of falling through to the Piper disk scan (which would
+  // mislabel them as Piper voices).
+  if (p === 'sapi' || p === 'windows-sapi') {
+    return WINDOWS_SAPI_VOICES.map(v => ({ id: v.id, gender: v.gender || '' }));
+  }
+  if (p === 'macos' || p === 'macos-say' || p === 'say') {
+    return MACOS_VOICES.map(v => ({ id: v.id, gender: v.gender || '' }));
+  }
+
   // Single-voice providers (e.g. soprano): one synthetic entry named after the
   // provider — every agent ends up sharing it, which is correct for these.
   if (isSingleVoiceProvider(p)) {
@@ -123,4 +102,4 @@ export function voicesForProvider(provider, { scanInstalledVoices, getVoiceMeta
   return [];
 }
 
-export default { ELEVENLABS_VOICES, KOKORO_VOICE_IDS, kokoroGender, voicesForProvider };
+export default { ELEVENLABS_VOICES, KOKORO_VOICE_IDS, WINDOWS_SAPI_VOICES, MACOS_VOICES, kokoroGender, voicesForProvider };

package/src/services/tts-engine-service.js

@@ -53,6 +53,22 @@ export function getAvailableEngines() {
   return TTS_ENGINES.filter(e => !e.platform || e.platform === process.platform);
 }
 
+/** Is this engine runnable on the current OS? (Native engines are platform-bound.) */
+export function isEngineSupported(engineId) {
+  const e = TTS_ENGINES.find(x => x.id === engineId);
+  return !!e && (!e.platform || e.platform === process.platform);
+}
+
+/**
+ * Every engine, each tagged with `supported` for the current platform. Unlike
+ * getAvailableEngines() (which hides off-platform engines), this lists them all
+ * so the picker can show e.g. Windows SAPI greyed "(Not supported)" on Linux
+ * instead of appearing to be missing.
+ */
+export function getAllEngines() {
+  return TTS_ENGINES.map(e => ({ ...e, supported: !e.platform || e.platform === process.platform }));
+}
+
 export function checkEngineInstalled(engineId) {
   const engine = TTS_ENGINES.find(e => e.id === engineId);
   if (!engine) return false;
@@ -110,6 +126,19 @@ export function checkEngineInstalled(engineId) {
   return false;
 }
 
+/**
+ * Map a picker engine id to the provider name the SSH-remote sender + receiver
+ * understand. play-tts-ssh-remote.sh validates
+ * piper|soprano|macos|windows-sapi|kokoro|elevenlabs, but the picker/config use
+ * 'sapi' and 'macos-say'. This is the single translation point.
+ * @param {string} engineId
+ * @returns {string} receiver/sender provider name
+ */
+const RECEIVER_PROVIDER_MAP = { sapi: 'windows-sapi', 'macos-say': 'macos' };
+export function receiverProviderId(engineId) {
+  return RECEIVER_PROVIDER_MAP[engineId] || engineId;
+}
+
 export function getEngineStatuses() {
   return getAvailableEngines().map(e => ({
     ...e,

package/src/utils/provider-validator.js

@@ -7,6 +7,64 @@ import { spawnSync } from 'node:child_process';
 import path from 'node:path'; // For safe path operations and traversal prevention
 import fs from 'node:fs'; // For checking file/directory existence
 import os from 'node:os'; // For os.homedir() to prevent HOME injection attacks
+import { listProviders, displayName as catalogDisplayName } from '../services/provider-catalog.js';
+
+/**
+ * Canonical set of TTS synthesis providers AgentVibes knows how to use.
+ * SINGLE SOURCE OF TRUTH: every user-facing dispatcher (MCP set_provider
+ * allowlist, voice switch, voice list, Windows provider manager) must recognise
+ * these providers — platform-scoped where noted below. A provider that lives
+ * here but is missing from a dispatcher is the "silent drop to Piper / Unknown
+ * provider" class of bug (see AVI-S8.1). The dispatcher-parity conformance test
+ * asserts this invariant.
+ *
+ * Platform notes: `macos` is Darwin-only; `windows-sapi`/`windows-piper` are
+ * Windows-only. `kokoro` ships a play-tts runtime on BOTH Unix
+ * (play-tts-kokoro.sh) and PowerShell (play-tts-kokoro.ps1). `elevenlabs` ships
+ * a Unix runtime only (play-tts-elevenlabs.sh) — there is NO
+ * play-tts-elevenlabs.ps1 yet — so it is recognised by the portable (bash/JS)
+ * dispatchers but intentionally excluded from the Windows provider manager
+ * until a PowerShell runtime lands (AVI-S8.2).
+ * @type {readonly string[]}
+ */
+export const SUPPORTED_PROVIDERS = Object.freeze(listProviders().map((p) => p.id));
+
+/**
+ * Providers that use the SAME id on every OS and have at least a Unix (bash)
+ * runtime. Every PORTABLE dispatcher — the MCP set_provider allowlist,
+ * voice-manager.sh, and list-voices.js — must recognise these regardless of
+ * platform. DERIVED VIEW over the catalog: the providers with a Unix runtime
+ * whose voices are enumerable without disk (static/name-to-id) — i.e. kokoro and
+ * elevenlabs.
+ * @type {readonly string[]}
+ */
+export const CROSS_PLATFORM_PROVIDERS = Object.freeze(
+  listProviders()
+    .filter((p) => p.runtime.unix !== null && (p.voiceModel === 'static' || p.voiceModel === 'name-to-id'))
+    .map((p) => p.id),
+);
+
+/**
+ * Subset of {@link CROSS_PLATFORM_PROVIDERS} that ALSO ships a PowerShell runtime
+ * (play-tts-*.ps1), so the Windows provider manager (provider-manager.ps1) must
+ * recognise them too. `elevenlabs` is excluded until play-tts-elevenlabs.ps1
+ * exists (AVI-S8.2). DERIVED VIEW over the catalog.
+ * @type {readonly string[]}
+ */
+export const WINDOWS_RUNTIME_PROVIDERS = Object.freeze(
+  listProviders()
+    .filter((p) => p.runtime.unix !== null && (p.voiceModel === 'static' || p.voiceModel === 'name-to-id') && p.runtime.windows !== null)
+    .map((p) => p.id),
+);
+
+/**
+ * Is `name` a provider AgentVibes canonically supports?
+ * @param {string} name
+ * @returns {boolean}
+ */
+export function isKnownProvider(name) {
+  return SUPPORTED_PROVIDERS.includes(String(name || '').toLowerCase());
+}
 
 /**
  * Helper: Check if command exists in PATH
@@ -603,16 +661,8 @@ function getPackageInfo(pkgName) {
  * @returns {string} Display name
  */
 export function getProviderDisplayName(providerName) {
-  const names = {
-    soprano: 'Soprano TTS',
-    piper: 'Piper TTS',
-    kokoro: 'Kokoro TTS',
-    elevenlabs: 'ElevenLabs',
-    macos: 'macOS Say',
-    sapi: 'Windows SAPI',
-    'windows-sapi': 'Windows SAPI',
-    'windows-piper': 'Piper TTS'
-  };
-
-  return names[providerName] || providerName;
+  // DERIVED VIEW: delegate to the Provider Catalog (single source of truth for
+  // display names). The catalog resolves aliases (e.g. 'sapi' → Windows SAPI) and
+  // passes unknown ids through unchanged, preserving this function's contract.
+  return catalogDisplayName(providerName);
 }

package/.claude/commands/agent-vibes/language.md

@@ -1,23 +0,0 @@
----
-description: Set your main/native language for learning mode
----
-
-Set your main/native language. This is the language you already know and will hear first when learning mode is enabled.
-
-Usage:
-```
-/agent-vibes:language english
-/agent-vibes:language spanish
-/agent-vibes:language french
-```
-
-The main language uses your currently selected voice. When learning mode is ON, TTS will speak in your main language FIRST, then translate to your target language.
-
-Default: english
-
-Supported languages: english, spanish, french, german, italian, portuguese, chinese, japanese, korean, hindi, arabic, polish, dutch, turkish, swedish, russian, and 15+ more.
-
-After setting your main language:
-1. Set your target language with `/agent-vibes:target <language>`
-2. Set target voice with `/agent-vibes:target-voice <voice>`
-3. Enable learning mode with `/agent-vibes:learn`

package/.claude/commands/agent-vibes/learn.md

@@ -1,67 +0,0 @@
----
-description: Enable or disable language learning mode
----
-
-Turn language learning mode ON or OFF. When enabled, Claude will speak acknowledgments and completions in BOTH your main language and target language.
-
-Usage:
-```
-/agent-vibes:learn              # Turn ON
-/agent-vibes:learn off          # Turn OFF
-/agent-vibes:learn status       # Show current setup
-```
-
-## How Learning Mode Works:
-
-When learning mode is **ON**:
-1. **First**: Speak in your main language (using your current voice)
-2. **Then**: Speak the SAME message translated to your target language (using target voice)
-
-Example:
-```
-Main language (English, Aria): "I'll check that for you"
-Target language (Spanish, Antoni): "Lo verificaré para ti"
-```
-
-## Setup Steps:
-
-1. Set your main language:
-   ```
-   /agent-vibes:language english
-   ```
-
-2. Set your target language:
-   ```
-   /agent-vibes:target spanish
-   ```
-
-3. Set target voice (recommended):
-   ```
-   /agent-vibes:target-voice Antoni
-   ```
-
-4. Enable learning mode:
-   ```
-   /agent-vibes:learn
-   ```
-
-5. Check your setup:
-   ```
-   /agent-vibes:learn status
-   ```
-
-## Notes:
-
-- Translations are **direct translations** of what was said in the main language
-- Same **personality/sentiment** applies to both languages
-- Works with all AgentVibes features (BMAD, personalities, etc.)
-- Requires multilingual voices for target language (Antoni, Rachel, Domi, Bella, etc.)
-- Small pause (0.5s) between main and target language announcements
-
-## Disable Learning Mode:
-
-```
-/agent-vibes:learn off
-```
-
-This returns to normal single-language TTS mode.

package/.claude/commands/agent-vibes/replay-target.md

@@ -1,14 +0,0 @@
----
-description: Replay the last target language audio (for language learning mode)
----
-
-Replay the last message that was spoken in your target language during language learning mode.
-
-This is useful when learning a new language - you can hear the translation again without triggering a new one.
-
-Usage:
-- `/agent-vibes:replay-target` - Replay the last target language audio
-
-**Note:** This only works when language learning mode is active (`/agent-vibes:learn`).
-
-!bash .claude/hooks/replay-target-audio.sh

package/.claude/commands/agent-vibes/target-voice.md

@@ -1,26 +0,0 @@
----
-description: Set the voice for your target language
----
-
-Set which voice to use when speaking your target language. This should typically be a multilingual voice that supports your target language.
-
-Usage:
-```
-/agent-vibes:target-voice Antoni
-/agent-vibes:target-voice Rachel
-/agent-vibes:target-voice Domi
-```
-
-Recommended multilingual voices:
-- **Antoni** - Best for Spanish, Portuguese
-- **Rachel** - Best for French, English
-- **Domi** - Best for German, European languages
-- **Bella** - Best for Italian, Romance languages
-- **Charlotte** - European languages
-- **Matilda** - Latin languages
-
-These voices support 30+ languages using Piper TTS' Multilingual v2 model.
-
-After setting your target voice:
-- Enable learning mode with `/agent-vibes:learn`
-- Check your setup with `/agent-vibes:learn status`

package/.claude/commands/agent-vibes/target.md

@@ -1,30 +0,0 @@
----
-description: Set the language you want to learn
----
-
-Set the target language you want to learn. When learning mode is enabled, TTS will speak in your main language FIRST, then speak the translation in your target language.
-
-Usage:
-```
-/agent-vibes:target spanish
-/agent-vibes:target french
-/agent-vibes:target german
-```
-
-Recommended voices by target language:
-- Spanish → Antoni (Piper TTS) / es_ES-davefx-medium (Piper)
-- French → Rachel (Piper TTS) / fr_FR-siwis-medium (Piper)
-- German → Domi (Piper TTS) / de_DE-thorsten-medium (Piper)
-- Italian → Bella (Piper TTS) / it_IT-riccardo-x_low (Piper)
-- Portuguese → Matilda (Piper TTS) / pt_BR-faber-medium (Piper)
-- Chinese → Amy (Piper TTS) / zh_CN-huayan-medium (Piper)
-- Japanese → Antoni (Piper TTS) / ja_JP-hikari-medium (Piper)
-- Other languages → Antoni (Piper TTS) / check available Piper voices
-
-**Note:** The system will automatically suggest the correct voice based on your active TTS provider. After setting your target language, the suggestion will match whether you're using Piper TTS or Piper.
-
-After setting your target language:
-1. Set the voice for target language with `/agent-vibes:target-voice <voice>`
-2. Enable learning mode with `/agent-vibes:learn`
-
-Supported languages: spanish, french, german, italian, portuguese, chinese, japanese, korean, hindi, arabic, polish, dutch, turkish, swedish, russian, and 15+ more.